MAROKO133 Hot ai: Mysterious Interstellar Object Has Sprouted a Tendril Reaching Toward th

📌 MAROKO133 Hot ai: Mysterious Interstellar Object Has Sprouted a Tendril Reaching

Interstellar object 3I/ATLAS continues to fascinate astronomers as it rips through our solar system. And the more we find out about the object — widely suspected to be an icy comet — the more questions emerge.

Latest among those mysteries: the Keck II telescope in Hawaii observed the object when it was just over 2.5 times the distance between the Earth and Sun back in August, and found “evidence for a puzzling anti-tail extension in the direction of the Sun,” as Harvard astronomer Avi Loeb wrote in a blog post last week.

In a recent paper that has yet to be peer reviewed, a team of astronomers used the Keck data to confirm “previously reported cyanide and nickel outgassing,” which are being emitted both in and against the direction of the Sun, which offers “clear evidence for an anti-tail,” according to Loeb.

“Most remarkably, the white light image of 3I/ATLAS does not show evidence for a familiar cometary tail, as expected for dust which scatters sunlight and is pushed away from the Sun by solar radiation pressure,” he added.

As IFLScience explains, the phenomenon could be the result of natural processes. One possibility is that it’s a type of optical illusion; because of the Earth’s relative location in space, a comet’s wide tail can fan out from behind it to make it look as if it has a tail growing tail sprouting from either side.

Another possibility is that larger grains of dust are refusing to be pushed away by solar wind on the comet’s Sun-facing side. The comet’s core of sublimating ice could be spinning rapidly and releasing large pieces of debris in both directions along its orbit making it appear to have a Sunward “anti-tail” in addition to its regular tail.

Scientists have previously identified other comets showing Sun-facing “anti-tails” that suggest the “slow ejection of relatively large dust particles predominantly from the sunlit hemisphere.”

“With a rotating comet nucleus… ejecta from a spot can come off with heliocentric velocity that puts it either in front of or behind the nucleus,” explained University of California, Los Angeles planetary astronomer Michael Busch in a post on Bluesky. “It does not matter which side it starts from.”

“Small dust and ejected gas gets pushed out by radiation pressure and solar wind,” he explained in a followup. “But larger pieces of ejecta spread out along the orbit; both in front of and behind the nucleus.”

To Loeb, the anti-tail remains an “anomaly that raises two questions,” according to a more recent blog post. “What is the nature of the anti-tail? Why are comet experts ignoring this anomaly while insisting that 3I/ATLAS is a familiar comet?”

Fortunately, before 3I/ATLAS leaves the solar system for good, it will provide us with several more opportunities to examine it. It’s expected to make a close approach Jupiter next month, giving NASA’s Juno spacecraft and the European Space Agency’s Juice spacecraft a chance to get a brief glimpse.

For now, Loeb ranks 3I/ATLAS as a four out of ten on his “Loeb scale” — which he invented to gauge the likeliness of an interstellar object being extraterrestrial technology — in a figure that he says means that it has “increasingly anomalous characteristics.”

More on 3I/ATLAS: Interstellar Object Is Spraying Something Weird, Scientists Find

The post Mysterious Interstellar Object Has Sprouted a Tendril Reaching Toward the Sun appeared first on Futurism.

🔗 Sumber: futurism.com

📌 MAROKO133 Update ai: MIT Researchers Unveil “SEAL”: A New Step Towards Self-Impr

The concept of AI self-improvement has been a hot topic in recent research circles, with a flurry of papers emerging and prominent figures like OpenAI CEO Sam Altman weighing in on the future of self-evolving intelligent systems. Now, a new paper from MIT, titled “Self-Adapting Language Models,” introduces SEAL (Self-Adapting LLMs), a novel framework that allows large language models (LLMs) to update their own weights. This development is seen as another significant step towards the realization of truly self-evolving AI.

The research paper, published yesterday, has already ignited considerable discussion, including on Hacker News. SEAL proposes a method where an LLM can generate its own training data through “self-editing” and subsequently update its weights based on new inputs. Crucially, this self-editing process is learned via reinforcement learning, with the reward mechanism tied to the updated model’s downstream performance.

The timing of this paper is particularly notable given the recent surge in interest surrounding AI self-evolution. Earlier this month, several other research efforts garnered attention, including Sakana AI and the University of British Columbia’s “Darwin-Gödel Machine (DGM),” CMU’s “Self-Rewarding Training (SRT),” Shanghai Jiao Tong University’s “MM-UPT” framework for continuous self-improvement in multimodal large models, and the “UI-Genie” self-improvement framework from The Chinese University of Hong Kong in collaboration with vivo.

Adding to the buzz, OpenAI CEO Sam Altman recently shared his vision of a future with self-improving AI and robots in his blog post, “The Gentle Singularity.” He posited that while the initial millions of humanoid robots would need traditional manufacturing, they would then be able to “operate the entire supply chain to build more robots, which can in turn build more chip fabrication facilities, data centers, and so on.” This was quickly followed by a tweet from @VraserX, claiming an OpenAI insider revealed the company was already running recursively self-improving AI internally, a claim that sparked widespread debate about its veracity.

Regardless of the specifics of internal OpenAI developments, the MIT paper on SEAL provides concrete evidence of AI’s progression towards self-evolution.

Understanding SEAL: Self-Adapting Language Models

The core idea behind SEAL is to enable language models to improve themselves when encountering new data by generating their own synthetic data and optimizing their parameters through self-editing. The model’s training objective is to directly generate these self-edits (SEs) using data provided within the model’s context.

The generation of these self-edits is learned through reinforcement learning. The model is rewarded when the generated self-edits, once applied, lead to improved performance on the target task. Therefore, SEAL can be conceptualized as an algorithm with two nested loops: an outer reinforcement learning (RL) loop that optimizes the generation of self-edits, and an inner update loop that uses the generated self-edits to update the model via gradient descent.

This method can be viewed as an instance of meta-learning, where the focus is on how to generate effective self-edits in a meta-learning fashion.

A General Framework

SEAL operates on a single task instance (C,τ), where C is context information relevant to the task, and τ defines the downstream evaluation for assessing the model’s adaptation. For example, in a knowledge integration task, C might be a passage to be integrated into the model’s internal knowledge, and τ a set of questions about that passage.

Given C, the model generates a self-edit SE, which then updates its parameters through supervised fine-tuning: θ′←SFT(θ,SE). Reinforcement learning is used to optimize this self-edit generation: the model performs an action (generates SE), receives a reward r based on LMθ′’s performance on τ, and updates its policy to maximize the expected reward.

The researchers found that traditional online policy methods like GRPO and PPO led to unstable training. They ultimately opted for ReST^EM, a simpler, filtering-based behavioral cloning approach from a DeepMind paper. This method can be viewed as an Expectation-Maximization (EM) process, where the E-step samples candidate outputs from the current model policy, and the M-step reinforces only those samples that yield a positive reward through supervised fine-tuning.

The paper also notes that while the current implementation uses a single model to generate and learn from self-edits, these roles could be separated in a “teacher-student” setup.

Instantiating SEAL in Specific Domains

The MIT team instantiated SEAL in two specific domains: knowledge integration and few-shot learning.

• Knowledge Integration: The goal here is to effectively integrate information from articles into the model’s weights.
• Few-Shot Learning: This involves the model adapting to new tasks with very few examples.

Experimental Results

The experimental results for both few-shot learning and knowledge integration demonstrate the effectiveness of the SEAL framework.

In few-shot learning, using a Llama-3.2-1B-Instruct model, SEAL significantly improved adaptation success rates, achieving 72.5% compared to 20% for models using basic self-edits without RL training, and 0% without adaptation. While still below “Oracle TTT” (an idealized baseline), this indicates substantial progress.

For knowledge integration, using a larger Qwen2.5-7B model to integrate new facts from SQuAD articles, SEAL consistently outperformed baseline methods. Training with synthetically generated data from the base Qwen-2.5-7B model already showed notable improvements, and subsequent reinforcement learning further boosted performance. The accuracy also showed rapid improvement over external RL iterations, often surpassing setups using GPT-4.1 generated data within just two iterations.

Qualitative examples from the paper illustrate how reinforcement learning leads to the generation of more detailed self-edits, resulting in improved performance.

While promising, the researchers also acknowledge some limitations of the SEAL framework, including aspects related to catastrophic forgetting, computational overhead, and context-dependent evaluation. These are discussed in detail in the original paper.

Original Paper: https://arxiv.org/pdf/2506.10943

Project Site: https://jyopari.github.io/posts/seal

Github Repo: https://github.com/Continual-Intelligence/SEAL

The post MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI first appeared on Synced.

🔗 Sumber: syncedreview.com